ASP.NET Core: Create and Open PDF (in browser)

Question

Working in ASP.NET Core and using iTextSharp, I'm building a PDF and save it to the local system. I now want to open that file in the browser but can't seem to make that work since I get a FileStream error in one try and nothing at all in another try.

My logic is in the controller below. I've replaced the unnecessary code with // region description. The important code (the things I tried) is inside the TODO: Open the file region.

Controller

    [HttpPost]
    public (JsonResult, IActionResult) CreatePDF([FromBody] ReportViewModel model)
    {
        try
        {

            // region Logic code
            using (System.IO.MemoryStream memoryStream = new System.IO.MemoryStream())
            {
                // Create the document
                iTextSharp.text.Document document = new iTextSharp.text.Document();
                // Place the document in the PDFWriter
                iTextSharp.text.pdf.PdfWriter PDFWriter =
                    iTextSharp.text.pdf.PdfWriter.GetInstance(document, memoryStream);

                // region Initialize Fonts

                // Open the document
                document.Open();

                // region Add content to document

                // Close the document
                document.Close();

                #region Create and Write the file
                // Create the directory
                string directory = $"{_settings.Value.ReportDirectory}\\";
                System.IO.Directory.CreateDirectory(System.IO.Path.GetDirectoryName(directory));

                // region Create the fileName

                // Combine the directory and the fileName
                directory = System.IO.Path.Combine(directory, fileName);

                // Create the file
                byte[] content = memoryStream.ToArray();
                using (System.IO.FileStream fs = System.IO.File.Create(directory))
                {
                    fs.Write(content, 0, (int)content.Length);
                }
                #endregion

                #region TODO: Open the file
                // TRY: File(Stream, type) => Newtonsoft.Json.JsonSerializationException:
                // Error getting value from 'ReadTimeout' on 'System.IO.FileStream'.
                // ---> System.InvalidOperationException: Timeouts are supported for this stream.
                System.IO.FileStream fileStream = new System.IO.FileStream(directory, System.IO.FileMode.Open);
                var returnPDF = File(fileStream, contentType: "application/pdf");

                // TRY: File(path, type) => Newtonsoft.Json.JsonSerializationException:
                // Error getting value from 'ReadTimeout' on 'System.IO.FileStream'.
                // ---> System.InvalidOperationException: Timeouts are supported for this stream.
                returnPDF = File(System.IO.File.OpenRead(directory), contentType: "application/pdf" );

                // TRY: File(byte[], type) => Nothing happened
                returnPDF = File(content, contentType: "apllication/pdf");
                #endregion

                return ( Json(new { isError = false }), returnPDF );
            }
        }
        catch (System.Exception ex)
        {
            Serilog.Log.Error($"ReportController.CreatePDF() - {ex}");
            return ( Json(new { isError = true, errorMessage = ex.Message }), null );
        }
    }

References to useful Stackoverflow answers

• Return PDF to the Browser using Asp.net core
• Open PDF in a new tab in browser
• Web api controller method giving exception while serializing the Stream object
• stream.ReadTimeout threw an exception of type System.InvalidOperationException

Answer 1

Instead of loading the file again from the directory you can just create a FileContentResult and pass the byte[] to it, that you already allocated.

return ( Json(new { isError = false }), new FileContentResult(content, "application/pdf"));

Keep in mind that the browser will not download the file this way. It will extract the response-stream but not download it because it is within your json response and therefore the response-content-type will be application/json.

You could instead just return an IActionResult and just do it like that:

return new FileContentResult(content, "application/pdf");

If you persist on having the information "isError", then you could provide an url additionally and instead of the fileresult. For that you can register the IActionContextAccessor class in your ConfigureServices(...) method.

services.AddSingleton<IActionContextAccessor, ActionContextAccessor>();

Provide a seperate route and use the IUrlHelperFactory to generate an url.

A sample can be found here.

This url can then be "clicked" programatically within the front-end when the response isError equals false.

Answer 2

Action should return a single result, be it PDF or JSON, so that the client can handle the response appropriately.

I've refactored the code to make it easier to read by extracting the major logic out into a separate function.

The primary focus of the action being on how the response is returned.

[HttpPost]
public IActionResult CreatePDF([FromBody] ReportViewModel model) {
    try {
        // region Logic code
        byte[] content = CreateFile(model);
        return File(content, contentType: "application/pdf");
    } catch (System.Exception ex) {
        Serilog.Log.Error($"ReportController.CreatePDF() - {ex}");
        return Json(new { isError = true, errorMessage = ex.Message });
    }
}

byte[] CreateFile(ReportViewModel model) {
    using (System.IO.MemoryStream memoryStream = new System.IO.MemoryStream()) {
        // Create the document
        iTextSharp.text.Document document = new iTextSharp.text.Document();
        // Place the document in the PDFWriter
        iTextSharp.text.pdf.PdfWriter PDFWriter =
            iTextSharp.text.pdf.PdfWriter.GetInstance(document, memoryStream);

        // region Initialize Fonts
        // Open the document
        document.Open();
        // region Add content to document
        // Close the document
        document.Close();
        #region Create and Write the file
        // Create the directory
        string directory = $"{_settings.Value.ReportDirectory}\\";            
        System.IO.Directory.CreateDirectory(System.IO.Path.GetDirectoryName(directory));
        // region Create the fileName
        // Combine the directory and the fileName
        directory = System.IO.Path.Combine(directory, fileName);
        // Create the file
        byte[] content = memoryStream.ToArray();
        using (System.IO.FileStream fs = System.IO.File.Create(directory)) {
            fs.Write(content, 0, (int)content.Length);
        }
        #endregion            
        return content;
    }    
}

Answer 3

This is slightly adapted from my working code, but should suit your purposes. I believe there's some room for improvement (Like properly disposing of the MemoryStream) but it's succinct, clean and best of all, it works.

There's no need for the extra serialization, or read / write to the file system if it's not a requirement.

Controller

[HttpPost]
public (JsonResult, IActionResult) CreatePDF([FromBody] ReportViewModel model)
{
    try
    {
        return ( Json(new { isError = false }), GetDocument() );
    }

    catch (System.Exception ex)
    {
        Serilog.Log.Error($"ReportController.CreatePDF() - {ex}");
        return (Json(new {isError = true, errorMessage = ex.Message}), null);
    }
}

public FileContentResult GetDocument()
{
    var fc = new MyPdfCreator();
    var document = fc.GetDocumentStream();
    return File(document.ToArray(), "application/pdf", "File Name.pdf");
}

MyPdfCreator

public class MyPdfCreator 
{
    public MemoryStream GetDocumentStream()
    {
        var ms = new MemoryStream();
        var doc = new Document();
        var writer = PdfWriter.GetInstance(doc, ms);
        writer.CloseStream = false;
        doc.Open();
        doc.Add(new Paragraph("Hello World"));
        doc.Close();
        writer.Close();
        return ms;
    }
}